more raster data

[nevrome]

I'm currently in the field in Poland and played around with 3d visualization of raster data (DEM, orthophoto, geomagnetic, etc.) with the R package rgl. To share my code on stackoverflow I wanted to use the AtlantGIS example datasets. Unfortunately the DEM in the area of the fake geomagentic diagram is fairly boring and an orthophoto is not available.

I suggest to add an artificial area dataset (3-4ha) on Atlantis with:

• a fake high resolution DEM (simulating lidar or SFM data) with an interesting, meaningful relief
• a fake orthophoto
• a meaningful geomagnetic diagram

The constructed example feature could be a destroyed burial mound. A structure like this with a stone foundation could be partially visible in all three of the above mentioned raster datasets. It's quite easy to construct an artificial DEM and for the geomagnetic diagram we could use modified real world data. No idea how to construct a good orthophoto though.

@kacebe What do you think? This is just an idea - I'm not even sure whether it fits into AtlantGIS. I guess you also had plans to add more raster data?

[kacebe]

First of all: That sounds fantastic!
In fact I've also produced additional datasets for several more scenarios, that need more documentation before I merge them into the master. The dataset lacks metadata almost completely which is not a good idea esp. in teaching contexts (I'll open another issue on that).
I'd propose we keep the general folder structure oriented on data model (raster, vector, table). The raster folder already has subfolders for dem and geophys. I'd add dsm and dtm following the definition summarised here .
I guess your datasets will be of ASCII-type except the fake orthoimage?
I'm not really happy about binary data on github and I have several large georeferenced maps to add. They might much better be hosted on zenodo and referenced in the teaching material (again, I'll open another issue on that topic).
From your perspective: What would be the best way for you to add the data? Individual pull request? Do you use a local branch? Should I change the folder-structure before you start to contribute?
Should we integrate your data from earlier R-workshops on that occasion? Presumably @dirkseidensticker will have to add to that issue.

[nevrome]

Why do you dislike binary data on github? If the individual files are not too big I don't see an advantage of outsourcing individual files to zenodo. I suggest to keep the rasters as geotiffs within the repository.

There's also no real need to store the rasters in an ASCII-format - that would just bloat them unnecessarily and I guess nobody will ever need gits row-wise version control to track the change within one raster in this context. Or do you prefer a human readable format for other reasons?

The folder structure is fine and I'm okay with contributing within an own fork with pull-requests to your master. So far I don't have anything to add. Maybe it's possible to get my hands on a (modified) real world dataset that mirrors the above mentioned scenario. I'm going to speak with some people.

[nevrome]

@chrinne offered to provide some raw data of the Closos project. There we would have orthophotos (also historical ones), geomagnetics diagrams and some additional 3d data. We could modify the data a little bit and change its meta data (location, timestamps etc.).

[nevrome]

I have now the OK to publish a decontextualized version of the initial dataset I had in mind. I don't have the data yet, but I guess I'll get my hands on it during the next month(s).

It contains overlapping raster data for a few hectares field with one or two Bronze Age burial mounds:

• geomagnetic diagrams
• orthophoto
• high resolution SFM elevation model

Several questions arise, @kacebe:

• Where exactly should I place the data on the island Atlantis? Which coordinate reference system is in use there?
• How to decontextualize the metadata beyond spatial information? Timestamps? User info? @dirkseidensticker & @crinne, I guess you may have some input here.
• How to store the data? GeoTIFs? If so - which setup (nT range etc.) is suitable for the geomagnetic diagram? Resolution?
• Should the data be artificially altered to be even more speaking or should we stick to the real assemblage?

I'll also have to talk with the responsible project leaders if and how they would like to be mentioned in this repository.

Maybe it's best to figure this out in a Skype conference when I have the data.